aod-foundation
Skill95 starsChanged 37 days agoApache-2.0View on GitHub
Guided post-init workshop that helps new AOD Kit adopters establish product vision and design identity. Two-part flow: Part 1 (Vision) asks 5 guided questions to populate product-vision.md; Part 2 (Design) browses archetypes to generate brand files (brand.md, tokens.css, anti-patterns.md). Supports --vision and --design flags for partial execution. Use when a developer invokes /aod.foundation after running make init.
In this repository30 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ~aod-blueprintSkill
- ~aod-bugfixSkill
- ~aod-buildSkill
- ~aod-defineSkill
- ~aod-deliverSkill
- ~aod-discoverSkill
- ~aod-kickstartSkill
- aod-lensSkill
- aod-orchestrateSkill
- ~aod-planSkill
- ~aod-project-planSkill
- ~aod-runSkill
- ~aod-scoreSkill
- ~aod-specSkill
- aod-stackSkill
- ~aod-statusSkill
- code-execution-helperSkill
- git-workflow-helperSkill
- kb-createSkill
- kb-querySkill
- root-cause-analyzerSkill
- securitySkill
- tachi-control-analysisSkill
- tachi-infographicsSkill
- tachi-orchestrationSkill
- tachi-report-assemblySkill
- tachi-risk-scoringSkill
- tachi-sharedSkill
- tachi-threat-reportingSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
63 rules. About 0 are common, 37 are distinctive, and 26 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈37
Distinctive · estimate
This author's own choices. Worth a look.
26
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 19
Run after make init to replace placeholder markers with real content.
- Very specific: line 27
If arguments contain --vision:
- Distinctive: line 33
If neither flag is present:
- Distinctive: line 35
If both flags are present:
- Very specific: line 36
Treat as run_both = true (equivalent to no flags)
- Very specific: line 46
Use this ordered fallback to determine {project-name}:
- Very specific: line 48
Repository directory name: Run basename $(git rev-parse --show-toplevel) — use this if it yields a non-empty value
- Very specific: line 49
CLAUDE.md PROJECT_NAME: Read CLAUDE.md, grep for a line matching PROJECT_NAME that does NOT contain {{PROJECT_NAME}} (the un-replaced …
- Distinctive: line 50
AskUserQuestion fallback: If neither source yields a usable name, ask: "What is your project name?" with a text input.
- Very specific: line 60
If the file does not exist: Set vision_state = "missing" — Part 1 will create from scratch
- Very specific: line 61
If the file contains [To be refined or [To be defined: Set vision_state = "placeholder" — Part 1 activates normally
- Very specific: line 62
If the file exists with no placeholder markers: Set vision_state = "populated" — Part 1 enters re-run mode
- Very specific: line 66
Check for brand directories under brands/:
- Very specific: line 69
If no brand directory found (only _example): Set design_state = "missing" — Part 2 activates normally
- Very specific: line 70
If a brand directory exists: Set design_state = "populated" — Part 2 enters re-run mode
- Very specific: line 93
If vision_only: Part 1 → completion summary → end
- Very specific: line 95
If run_both: Part 1 → "Part 1 complete.
- Very specific: line 103
Part 2: only brands/{project_name}/ directory (brand.md, tokens.css, anti-patterns.md, reference/)
- Distinctive: line 122
After generating any file content, display it for user review before writing to disk.
- Distinctive: line 125
Use AskUserQuestion with options:
- Distinctive: line 129
Only write to disk on explicit "Write to file" confirmation
- Distinctive: line 130
If "Edit a section": loop back to step 1 after applying changes
- Distinctive: line 134
When existing populated content is detected (vision_state = "populated" or design_state = "populated"):
- Distinctive: line 141
If "Keep current": proceed to next part without modifying files
- Distinctive: line 142
If "Update specific sections": present only the selected sections for re-answering, merge with existing content
- Distinctive: line 143
If "Replace entirely": run the normal first-run flow
- Distinctive: line 157
Default suggestion: Derive from CLAUDE.md description if available, otherwise: "Developers and small teams building software products"
- Distinctive: line 164
Default suggestion: "They lack a structured approach to [domain]"
- Distinctive: line 171
Default suggestion: Derive from CLAUDE.md description if available
- Distinctive: line 178
Default suggestion: "It takes a [specific approach] that existing solutions don't"
- Distinctive: line 185
Default suggestion: "User adoption, engagement frequency, and [domain-specific outcome]"
- Distinctive: line 201
If all 5 questions are skipped: display warning — "All questions were skipped.
- Distinctive: line 205
After all 5 questions are answered (or skipped), generate the product-vision.md content.
- Very specific: line 271
When running the full workshop (run_both = true) and vision_state is "placeholder" or "missing":
- Distinctive: line 273
After displaying the state summary (Step 1d), before starting questions, offer:
- Distinctive: line 274
Use AskUserQuestion: "Ready to start the Vision Workshop (5 questions)?"
- Very specific: line 278
When vision_only = true: do NOT offer skip — proceed directly to questions.
- Very specific: line 282
When vision_state = "populated" (no placeholder markers):
- Very specific: line 284
Read existing product-vision.md and extract current values for each section
- Distinctive: line 296
If "Update specific sections": present checkboxes for which sections to update, then ask only the corresponding questions
- Distinctive: line 297
If "Replace entirely": run the full 5-question flow, overwriting the entire file
- Very specific: line 305
When design_state = "populated" (brand directory already exists):
- Very specific: line 307
Read existing brands/{project_name}/brand.md and extract: archetype name (from tagline or heading), heading font, body font, primary color
- Distinctive: line 317
If "Keep current": skip Part 2 entirely, preserve all brand files
- Distinctive: line 318
If "Update specific sections": allow modifying individual brand file sections, merge with existing
- Distinctive: line 319
If "Replace entirely": run the full archetype selection flow below
- Very specific: line 326
If no files found: display warning — "No design archetypes found in .claude/design/archetypes/.
- Distinctive: line 348
If numbered option: store the archetype name, read the full archetype file for value extraction
- Distinctive: line 349
If "Custom": proceed to Step 2c
- Distinctive: line 353
When the user selects "Custom" (no archetype):
- Distinctive: line 355
Ask: "What is your primary brand color?
- Distinctive: line 359
Ask: "What font style fits your brand?
- Distinctive: line 363
Ask: "What is the overall visual mood?
- Very specific: line 371
Create directory brands/{project_name}/ (if not exists)
- Very specific: line 372
Create empty brands/{project_name}/reference/ directory
- Distinctive: line 392
Read the selected archetype's sections:
- Distinctive: line 476
Read the file and extract the active pack name
- Distinctive: line 479
If no stack pack is active: skip silently.
- Distinctive: line 483
After generating all three files:
- Distinctive: line 502
Only write to disk on explicit "Write all files" confirmation
- Very specific: line 509
Missing archetypes directory: If .claude/design/archetypes/ has no .md files, warn and offer only the Custom design option (Step 2c)
- Very specific: line 510
Missing product-vision.md: When vision_state = "missing", create from scratch using the full 5-question flow — do not read or parse a …
- Distinctive: line 519
After all parts complete, display a summary table:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
/aod.foundation Skill
Purpose
Guided post-init workshop with two parts:
- Part 1 (Vision): 5 storytelling-arc questions that populate
product-vision.md - Part 2 (Design): Archetype browser that generates
brands/{project-name}/files
Run after make init to replace placeholder markers with real content.
Step 0: Parse Flags
Parse --vision and --design flags from the user's arguments.
- If arguments contain
--vision:- Set
vision_only = true - Strip
--visionfrom arguments
- Set
- If arguments contain
--design:- Set
design_only = true - Strip
--designfrom arguments
- Set
- If neither flag is present:
- Set
run_both = true(full workshop)
- Set
- If both flags are present:
- Treat as
run_both = true(equivalent to no flags)
- Treat as
Step 1: State Detection
Detect the current project state to determine which parts to activate and whether this is a first-run or re-run.
1a: Derive Project Name
Use this ordered fallback to determine {project-name}:
- Repository directory name: Run
basename $(git rev-parse --show-toplevel)— use this if it yields a non-empty value - CLAUDE.md PROJECT_NAME: Read
CLAUDE.md, grep for a line matchingPROJECT_NAMEthat does NOT contain{{PROJECT_NAME}}(the un-replaced placeholder). Extract the value. - AskUserQuestion fallback: If neither source yields a usable name, ask: "What is your project name?" with a text input.
Sanitize the derived name to kebab-case: lowercase, replace spaces/underscores with hyphens, strip non-alphanumeric/non-hyphen characters.
Store as project_name for use throughout the workshop.
1b: Check Vision State
Read docs/product/01_Product_Vision/product-vision.md:
- If the file does not exist: Set
vision_state = "missing"— Part 1 will create from scratch - If the file contains
[To be refinedor[To be defined: Setvision_state = "placeholder"— Part 1 activates normally - If the file exists with no placeholder markers: Set
vision_state = "populated"— Part 1 enters re-run mode
1c: Check Design State
Check for brand directories under brands/:
- Glob
brands/*/brand.mdexcludingbrands/_example/brand.md - If no brand directory found (only
_example): Setdesign_state = "missing"— Part 2 activates normally - If a brand directory exists: Set
design_state = "populated"— Part 2 enters re-run mode
1d: Determine Flow
| vision_only | design_only | run_both | Action |
| true | false | false | Execute Part 1 only, skip Part 2 |
| false | true | false | Execute Part 2 only, skip Part 1 |
| false | false | true | Execute Part 1, then Part 2 |
Display state summary:
Foundation Workshop
Project: {project_name}
Vision: {vision_state} → {Part 1 will: create/update/offer re-edit}
Design: {design_state} → {Part 2 will: create/offer re-edit}
Mode: {Full workshop | Vision only | Design only}1e: Execution Flow
Based on flag state, execute parts in order:
- If
vision_only: Part 1 → completion summary → end - If
design_only: Part 2 → completion summary → end - If
run_both: Part 1 →"Part 1 complete. Moving to Design Identity."→ Part 2 → completion summary → end
At each part entry, check the state for re-run handling. If user chooses "Keep current" during re-run detection, skip that part and advance to the next step in the flow. Choosing "Keep current" for Part 1 proceeds directly to Part 2 (if run_both). Choosing "Keep current" for Part 2 proceeds to the completion summary.
1f: Modification Isolation
Each part writes to independent file paths:
- Part 1: only
docs/product/01_Product_Vision/product-vision.md - Part 2: only
brands/{project_name}/directory (brand.md, tokens.css, anti-patterns.md, reference/)
Modifying vision leaves brand files untouched. Modifying design leaves product-vision.md untouched. "Keep current" means zero writes to that part's files.
Shared Patterns
Progress Announcements
Display progress at every transition point. Format:
- During vision questions:
"Vision: Question {N}/5 — {question_title}" - Between parts:
"Part 1 complete. Moving to Design Identity." - During design:
"Design: {step_description}" - At completion:
"Workshop complete. {N} file(s) created/updated."
Show-Before-Write
After generating any file content, display it for user review before writing to disk.
- Display the generated content (full file or relevant section)
- Use AskUserQuestion with options:
- "Write to file" — Write the content to disk, continue
- "Edit a section" — Ask which section to change, apply edit, re-display
- "Cancel" — Skip writing this file, continue to next step
- Only write to disk on explicit "Write to file" confirmation
- If "Edit a section": loop back to step 1 after applying changes
Re-Run Detection
When existing populated content is detected (vision_state = "populated" or design_state = "populated"):
- Display current content summary (key values, not full file)
- Use AskUserQuestion with options:
- "Keep current" — Skip this part entirely, preserve existing content
- "Update specific sections" — Show sections, let user pick which to re-answer
- "Replace entirely" — Run the full creation flow from scratch
- If "Keep current": proceed to next part without modifying files
- If "Update specific sections": present only the selected sections for re-answering, merge with existing content
- If "Replace entirely": run the normal first-run flow
Part 1: Vision Workshop
Vision Questions
Present these 5 questions sequentially, following a storytelling arc from audience to measurement.
Question 1 — "Who is this for?"
- Arc position: Setting the scene
- Prompt: "Who is the primary user of your product? Describe them in 1-2 sentences — their role, context, and what makes them your target audience."
- Example: "Spotify targets music listeners who want personalized discovery without manual playlist curation."
- Default suggestion: Derive from
CLAUDE.mddescription if available, otherwise: "Developers and small teams building software products" - Maps to: Target Users section
Question 2 — "What problem do they have?"
- Arc position: The conflict
- Prompt: "What specific problem or frustration does your target user face today? What are they struggling with or missing?"
- Example: "Notion users struggled with scattered notes, docs, and project boards across multiple disconnected tools."
- Default suggestion: "They lack a structured approach to [domain]"
- Maps to: Problem Statement section
Question 3 — "What does your product do?"
- Arc position: The solution
- Prompt: "In 1-2 sentences, what does your product do to solve that problem? Focus on the core action, not features."
- Example: "Linear provides a streamlined issue tracker that's fast enough to feel invisible, so engineering teams focus on building, not managing tools."
- Default suggestion: Derive from
CLAUDE.mddescription if available - Maps to: Core Value Proposition + Mission Statement + Core Capabilities
Question 4 — "How is it different?"
- Arc position: The twist
- Prompt: "What makes your approach unique? How is it different from alternatives or the status quo?"
- Example: "Unlike traditional CRMs, HubSpot started free and grew with the customer — the product itself was the sales pitch."
- Default suggestion: "It takes a [specific approach] that existing solutions don't"
- Maps to: Vision Statement
Question 5 — "How will you know it's working?"
- Arc position: The resolution
- Prompt: "What 2-3 metrics or signals would tell you the product is succeeding? Think about what you'd check after launch."
- Example: "Figma tracks multiplayer sessions (collaboration), time-to-first-design (activation), and team plan upgrades (revenue)."
- Default suggestion: "User adoption, engagement frequency, and [domain-specific outcome]"
- Maps to: Success Metrics
Question Presentation Flow
For each of the 5 questions:
- Display progress:
"Vision: Question {N}/5 — {question_title}" - Display the question prompt text
- Display the example (italicized)
- Use AskUserQuestion with:
- A text input for the user's answer
- A "Skip" option that marks the field as
[To be defined]
- Validate: if answer is empty and not skipped, re-prompt once
- Store the answer for document generation
If all 5 questions are skipped: display warning — "All questions were skipped. The vision file will contain only placeholder markers. Downstream governance (/aod.define) may flag incomplete vision."
Vision Document Generation
After all 5 questions are answered (or skipped), generate the product-vision.md content.
Section mapping (answer → document section):
| Answer | Document Section | Generation Rule |
| Q1 (Who) | Target Users | Use answer directly. If skipped: [To be defined] |
| Q2 (Problem) | Problem Statement | Use answer directly. If skipped: [To be defined] |
| Q3 (What) | Mission Statement | Condense to 1 sentence if longer. If skipped: [To be defined] |
| Q3 (What) | Core Value Proposition | Use full answer. If skipped: [To be defined] |
| Q3 (What) | Core Capabilities | Extract key capabilities as a table. If skipped: [To be defined] |
| Q4 (Different) | Vision Statement | Frame as forward-looking "In a world where..." narrative. If skipped: [To be defined] |
| Q5 (Metrics) | Success Metrics | Format as a table with Metric and Target columns. If skipped: [To be defined] |
Document structure — follow the existing product-vision.md format:
# Product Vision — {project_name}
**Last Updated**: {YYYY-MM-DD}
**Owner**: Product Manager (product-manager)
**Status**: Active
---
## Mission Statement
{derived from Q3}
## Vision Statement
{derived from Q4}
## Problem Statement
{derived from Q2}
## Core Value Proposition
{derived from Q3}
## Core Capabilities
| Capability | Description |
|---|---|
{derived from Q3 — extract 2-4 key capabilities}
## Target Users
### Primary: {derived from Q1}
{expand Q1 answer into user persona description}
## Success Metrics
| Metric | Target |
|---|---|
{derived from Q5 — format each metric as a row}
---
**Maintained By**: Product Manager (product-manager)After generation:
- Display the full generated document
- Apply show-before-write pattern
- Write to
docs/product/01_Product_Vision/product-vision.md
Skip Part 1 Entirely
When running the full workshop (run_both = true) and vision_state is "placeholder" or "missing":
After displaying the state summary (Step 1d), before starting questions, offer:
- Use AskUserQuestion: "Ready to start the Vision Workshop (5 questions)?"
- "Start" — Begin Question 1
- "Skip to Design" — Skip Part 1 entirely, proceed to Part 2 without modifying
product-vision.md
When vision_only = true: do NOT offer skip — proceed directly to questions.
Vision Re-Run Handling
When vision_state = "populated" (no placeholder markers):
- Read existing
product-vision.mdand extract current values for each section - Display summary:
Current Vision: Mission: {first 80 chars}... Vision: {first 80 chars}... Problem: {first 80 chars}... Value Prop: {first 80 chars}... Users: {first 80 chars}... Metrics: {count} defined - Apply re-run detection pattern (Keep current / Update specific sections / Replace entirely)
- If "Update specific sections": present checkboxes for which sections to update, then ask only the corresponding questions
- If "Replace entirely": run the full 5-question flow, overwriting the entire file
Part 2: Design Identity Workshop
Design Re-Run Handling
When design_state = "populated" (brand directory already exists):
- Read existing
brands/{project_name}/brand.mdand extract: archetype name (from tagline or heading), heading font, body font, primary color - Display summary:
Current Design Identity: Archetype: {name or "Custom"} Fonts: {heading font} + {body font} Primary: {color value} Temperature: {warm/cool/neutral} - Apply re-run detection pattern (Keep current / Update specific sections / Replace entirely)
- If "Keep current": skip Part 2 entirely, preserve all brand files
- If "Update specific sections": allow modifying individual brand file sections, merge with existing
- If "Replace entirely": run the full archetype selection flow below
Step 2a: Archetype Discovery
Discover available archetypes at runtime:
- Glob
.claude/design/archetypes/*.md - If no files found: display warning — "No design archetypes found in
.claude/design/archetypes/. Check your installation." Offer only the Custom option (skip to Step 2c). - For each archetype file, parse:
- Name: Derive from filename, title-cased (e.g.,
precision.md→ "Precision") - Tagline: First blockquote line (
> ...) after the# {Name} Archetypeheading - Font pairing: Extract Headings and Body font names from the "Font Pairing" table's Font column
- Color temperature: Extract from the
**Temperature**:line under "Color Palette Strategy"
- Name: Derive from filename, title-cased (e.g.,
Step 2b: Archetype Browser
Present discovered archetypes as a numbered selection list via AskUserQuestion.
Display "Design: Archetype Selection" as progress announcement.
Format per archetype:
{N}. {Name} — "{Tagline (first ~60 chars)}"
Fonts: {Heading Font} + {Body Font} | Temperature: {temperature}AskUserQuestion options: numbered options "1" through "{N}" for each archetype, plus "Custom".
After selection:
- If numbered option: store the archetype name, read the full archetype file for value extraction
- If "Custom": proceed to Step 2c
Step 2c: Custom Design Flow
When the user selects "Custom" (no archetype):
- Ask: "What is your primary brand color? (e.g., 'deep blue', '#1a365d', 'warm terracotta')"
- AskUserQuestion with text input
- Store as
custom_primary_color
- Ask: "What font style fits your brand? (e.g., 'clean and modern', 'editorial serif', 'technical monospace')"
- AskUserQuestion with text input
- Store as
custom_font_style
- Ask: "What is the overall visual mood? (e.g., 'professional and minimal', 'warm and inviting', 'bold and energetic')"
- AskUserQuestion with text input
- Store as
custom_mood
Generate brand files using these custom inputs. Select appropriate fonts, colors, radius, and shadow values that match the described mood and font style. Use design knowledge to produce a coherent set of tokens.
Step 2d: Brand Directory and brand.md Generation
- Create directory
brands/{project_name}/(if not exists) - Create empty
brands/{project_name}/reference/directory Generate
brand.mdfollowing the structure frombrands/_example/brand.md:Replacements from archetype (or custom inputs):
{{BRAND_NAME}}→{project_name}in title case{{BRAND_TAGLINE}}→ archetype tagline (or custom mood description)- Primary Colors table → from archetype's
--color-primary,--color-primary-light,--color-accentvalues with hex + oklch - Secondary Colors table → from archetype's
--color-secondary,--color-accent, and a neutral derived from palette - Semantic Colors table → from archetype's success/warning/error/info values
{{HEADING_FONT}}→ archetype heading font name{{BODY_FONT}}→ archetype body font name{{MONO_FONT}}→JetBrains Mono(default for all)- Personality / Voice → derive from archetype description and character notes
- Visual Tone → derive from archetype's Layout Philosophy and overall description
- Usage Notes → adapt from archetype's guidelines and constraints
Step 2e: Anti-Patterns Generation
Generate brands/{project_name}/anti-patterns.md from archetype constraints:
- Read the selected archetype's sections:
- "Usage Notes" → "Avoid {archetype} for" items and accessibility considerations
- "Shadow Depth" → guidelines (e.g., "Avoid lg shadows" for precision)
- "Border Radius" → guidelines (e.g., "Never exceed --radius-lg" for precision, "Sharp corners never used" for warmth)
- "Motion Style" → guidelines (e.g., "Avoid bounce effects" for precision)
- "Layout Philosophy" → patterns and constraints
- Generate anti-patterns following
brands/_example/anti-patterns.mdstructure with these categories:- Visual Constraints: From radius, shadow, and spacing guidelines
- Typography Constraints: From font pairing rationale and weight guidance
- Motion / Animation Constraints: From motion style guidelines
- Layout Constraints: From layout philosophy
- Color Constraints: From palette approach and accessibility notes
- Each constraint MUST be specific and actionable:
- Good: "Avoid border-radius larger than 6px on rectangular elements" (precision)
- Good: "Never use sharp corners on any element" (warmth)
- Bad: "Keep it clean" (not actionable)
Step 2f: Token Mapping and tokens.css Generation
Generate brands/{project_name}/tokens.css with all 19 semantic tokens.
- Read the canonical mapping from
.claude/skills/aod-foundation/references/token-mapping.md - Extract values from the selected archetype's markdown tables:
- Color Palette Strategy table → color values (hsl/oklch format)
- Font Pairing table → font family names and fallbacks
- Border Radius table → radius values
- Shadow Depth table → shadow values
- Apply rename layer (archetype name → tokens.css name):
--color-error→--color-destructive--color-surface→--color-background(and also informs--color-muted)- Body font →
--font-sans - Heading font →
--font-heading
- Apply derivation rules for missing tokens:
--color-secondaryif absent: primary hue + 30 degrees--color-mutedif absent: primary at 10% saturation, 95% lightness--shadow-lgif absent (precision, technical): derive from--shadow-mdwith 2.5x spread/offset, 1.25x opacity--radius-lgif absent:--radius-md* 1.5--font-mono: always"JetBrains Mono", "Fira Code", monospace
- Generate single-layer
:root {}block:
/**
* Brand Tokens — {project_name}
* Generated by /aod.foundation from {archetype_name} archetype
*/
:root {
/* --- Colors --- */
--color-primary: {value};
--color-secondary: {value};
--color-accent: {value};
--color-muted: {value};
--color-destructive: {value};
--color-success: {value};
--color-warning: {value};
--color-info: {value};
--color-background: {value};
--color-foreground: {value};
/* --- Typography --- */
--font-sans: "{body font}", {fallbacks};
--font-heading: "{heading font}", {fallbacks};
--font-mono: "JetBrains Mono", "Fira Code", monospace;
/* --- Border Radius --- */
--radius-sm: {value};
--radius-md: {value};
--radius-lg: {value};
/* --- Shadows --- */
--shadow-sm: {value};
--shadow-md: {value};
--shadow-lg: {value};
}Total: 19 individual tokens (10 color + 3 font + 3 radius + 3 shadow).
Step 2g: Stack Pack Compatibility Check
If .aod/stack-active.json exists:
- Read the file and extract the active pack name
- Display note:
"Note: Stack pack '{pack_name}' is active. Generated tokens use standard CSS custom property names compatible with the pack's scaffold."
If no stack pack is active: skip silently.
Step 2h: Preview and Write All Brand Files
After generating all three files:
- Display each file's content with progress announcements:
Design: Preview — brand.md {full content} Design: Preview — tokens.css {full content} Design: Preview — anti-patterns.md {full content} - Apply show-before-write pattern for ALL files together:
- "Write all files" — Write all three to
brands/{project_name}/ - "Edit a section" — Ask which file and section to change, apply edit, re-display
- "Cancel" — Skip writing brand files entirely
- "Write all files" — Write all three to
- Only write to disk on explicit "Write all files" confirmation
- After writing:
"Design: Brand files written to brands/{project_name}/"
Edge Cases
- Missing archetypes directory: If
.claude/design/archetypes/has no.mdfiles, warn and offer only the Custom design option (Step 2c) - Missing product-vision.md: When
vision_state = "missing", create from scratch using the full 5-question flow — do not read or parse a nonexistent file - Special characters in project name: Step 1a sanitization handles this — always use the kebab-case
project_namefor directory and file paths - All vision questions skipped: Warning displayed per Question Presentation Flow. Still proceed to Part 2 if
run_both = true— skipping vision does not block design - All of Part 2 skipped during full workshop: Display completion summary with design artifacts marked "Skipped"
Workshop Completion Summary
After all parts complete, display a summary table:
Foundation Workshop Complete
| Artifact | Path | Status |
|----------|------|--------|
| Product Vision | docs/product/01_Product_Vision/product-vision.md | {Created/Updated/Skipped} |
| Brand Identity | brands/{project-name}/brand.md | {Created/Updated/Skipped} |
| Design Tokens | brands/{project-name}/tokens.css | {Created/Updated/Skipped} |
| Anti-Patterns | brands/{project-name}/anti-patterns.md | {Created/Updated/Skipped} |
{count} file(s) created, {count} file(s) updated, {count} file(s) skipped.Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

